compare_matching {MultiObjMatch}

Generate covariate balance in different matches

Description

This is a wrapper function for use in evaluating covariate balance across different matches. It only works for 'Basic' version of matching (using dist_bal_match).

Usage

compare_matching(
  matching_result,
  cov_list = NULL,
  display_all = TRUE,
  stat = "mean.diff"
)

Arguments

matching_result

an object returned by the main matching function dist_bal_match

cov_list

(optional) factor of names of covariates that we want to evaluate covariate balance on; default is NULL. When set to NULL, the program will compare the covariates that have been used to construct a propensity model.

display_all

(optional) boolean value of whether to display all the matches; default is TRUE, where matches at each quantile is displayed

stat

(optional) character of the name of the statistic used for measuring covariate balance; default is "mean.diff". This argument is the same as used in "cobalt" package, see: bal.tab

Value

a dataframe that shows covariate balance in different matches

Examples

## Generate matches 
data("lalonde", package="cobalt")
ps_cols <- c("age", "educ", "married", "nodegree", "race")
treat_val <- "treat"
response_val <- "re78"  
pair_dist_val <- c("age", "married", "educ", "nodegree", "race")
my_bal_val <- c("race")
r1s <- c(0.01,1,2,4,4.4,5.2,5.4,5.6,5.8,6)
r2s <- c(0.001)
match_result <- dist_bal_match(data=lalonde, treat_col= treat_val, 
marg_bal_col = my_bal_val, exclusion_penalty=r1s, balance_penalty=r2s, 
dist_col = pair_dist_val, 
propensity_col = ps_cols, max_iter=0)

## Generate table for comparing matches
compare_matching(match_result, display_all = TRUE)